As the annual average temperature continues to rise due to climate change caused by global warming, the incidence of heat diseases and the number of deaths are also increasing, which is expected to require various alternatives and research. In this study, the average temperature rise-related variables are extracted through statistical analysis for Wonju City, where the average temperature increase rate and change are high, and the average temperature is predicted by utilizing deep learning-based LSTM and GRU based on the extracted variables. Three models were extracted through correlation and regression analysis for 26 variables collected based on prior research consideration, based on which LSTM and GRU analysis were conducted. The analysis showed the lowest MSE of LSTM – 0.4399(2.94°C), GRU – 0.4444(2.97°C) in the third model with 12 variables, with little MAE difference between validation and test data. This study is significant in that it extracted variables through statistical analysis and predicted average temperature rise using deep learning as a data acquisition method for adapting the annual average temperature rise problem. In addition, it is expected that urban space factors that affect the average temperature rise in Wonju City will be extracted along with predicting the trend of average temperature change, and appropriate measures will be prepared to take into account regional impact factors, not uniform climate change adaptation.